/// <summary>
/// 根据C9Article 抓取视频数据
/// </summary>
/// <param name="skip">领衔量</param>
/// <param name="number">数量</param>
/// <returns></returns>
public async Task <List <C9Videos> > SaveVideosAsync(int skip = 0, int number = 100)
{
Console.WriteLine($"start:{skip}");
var C9Articles = Context.C9Articles
.OrderByDescending(m => m.UpdatedTime)
.Skip(skip).Take(number).ToList();
var videoList = new List <C9Videos>();
var lastVideo = Context.C9Videos.OrderByDescending(m => m.UpdatedTime).Take(60).ToList();
Parallel.ForEach(C9Articles, a =>
{
// 过滤非视频数据
if (a.Duration == null)
{
Console.WriteLine("Not Video" + a.Title);
return;
}
// 数据库去重
if (lastVideo.Any(m => m.SourceUrl.Equals(a.SourceUrl)))
{
return;
}
C9Videos re = _helper.GetPageVideo(a).Result;
if (re != null)
{
re.Id = Guid.NewGuid();
videoList.Add(re);
}
});
try
{
Context.AddRange(videoList);
await Context.SaveChangesAsync();
return(videoList);
}
catch (Exception e)
{
Log.Write("C9VideoSaveError.txt", e.Message);
Console.WriteLine(e);
return(null);
}
}
/// <summary>
/// 根据url抓取video内容
/// </summary>
public void SaveVideoByUrl()
{
var file = new FileInfo("c9videoGetErrors.txt");
StreamReader stream = file.OpenText();
string url = stream.ReadLine();
int i = 1;
while (!stream.EndOfStream)
{
if (!IsNullOrEmpty(url))
{
url = url.Trim();
C9Videos re = _helper.GetPageVideoByUrl(url);
if (Context.C9Videos.Any(m => m.SourceUrl == re.SourceUrl))
{
url = stream.ReadLine();
continue;
}
re.Id = Guid.NewGuid();
Context.C9Videos.Add(re);
try
{
Context.SaveChanges();
Console.WriteLine($"strat:{i}");
i++;
}
catch (Exception)
{
Log.Write("C9Video.error.txt", url);
}
}
url = stream.ReadLine();
}
Console.WriteLine("Done");
}
/// <summary>
/// 抓取单页视频内容
/// </summary>
/// <param name="article"></param>
/// <returns></returns>
public async Task <C9Videos> GetPageVideo(C9Articles article)
{
var video = new C9Videos
{
Duration = article.Duration,
SeriesTitle = article.SeriesTitle,
SeriesTitleUrl = article.SeriesTitleUrl,
SourceUrl = article.SourceUrl,
Title = article.Title,
ThumbnailUrl = article.ThumbnailUrl
};
if (article.SeriesTitleUrl != null)
{
video.SeriesType = article.SeriesTitleUrl.Substring(1);
video.SeriesType = video.SeriesType.Substring(0, video.SeriesType.IndexOf(@"/"));
}
string url = C9Daemon + article.SourceUrl;
try
{
var hw = new HtmlWeb();
// option获取InnerText，需要加以下设置
HtmlAgilityPack.HtmlNode.ElementsFlags.Remove("option");
HtmlDocument htmlDoc = await hw.LoadFromWebAsync(url);
HtmlNode mainNode = htmlDoc.DocumentNode.SelectSingleNode(".//main[@role='main']");
video.Author = mainNode.SelectSingleNode(".//div[@class='authors']")?.Descendants("a")?.Select(s => s.InnerText)
.ToArray().Join();
video.Language = mainNode.SelectSingleNode(".//div[@class='itemHead holder' and @dir='ltr']")?
.GetAttributeValue("lang", Empty);
video.Description = mainNode.SelectSingleNode(".//section[@class='ch9tab description']/div[@class='ch9tabContent']")?
.InnerText;
video.VideoEmbed = mainNode.SelectSingleNode(".//section[@class='ch9tab embed']/div[@class='ch9tabContent']")?
.InnerHtml;
var downloadUrls = mainNode.SelectNodes(".//section[@class='ch9tab download']//div[@class='download']//ul//li")?
.Select(s => new
{
text = s.Element("a").Attributes["download"]?.Value,
value = s.Element("a").Attributes["href"]?.Value
}).ToList();
if (downloadUrls != null)
{
foreach (var downloadUrl in downloadUrls)
{
var downloadType = downloadUrl.text?.ToLower().Trim();
if (string.IsNullOrEmpty(downloadType))
{
continue;
}
if (downloadType.Contains(".mp3"))
{
video.Mp3Url = downloadUrl.value;
}
else if (downloadType.Contains("low.mp4"))
{
video.Mp4LowUrl = downloadUrl.value;
}
else if (downloadType.Contains("mid.mp4"))
{
video.Mp4MidUrl = downloadUrl.value;
}
else if (downloadType.Contains("high.mp4"))
{
video.Mp4HigUrl = downloadUrl.value;
}
}
}
video.Tags = mainNode
.SelectNodes(".//section[@class='ch9tab description']//div[@class='ch9tabContent']//div[@class='tags']//a")?
.Select(s => s.InnerText).ToArray()?.Join();
video.Views = 0;
//获取日期 结构不同时处理
var date = mainNode.SelectSingleNode(".//time[@class='timeHelper']")?
.GetAttributeValue("datetime", Empty);
if (date == null)
{
//event
date = mainNode.SelectSingleNode(".//div[@class='releaseDate']")?.InnerText;
Console.WriteLine(date);
date = date.Replace("at", "");
video.CreatedTime = DateTime.ParseExact(date.Trim(), "MMM dd, yyyy h:mmtt", CultureInfo.CreateSpecificCulture("en-US"));
}
else
{
video.CreatedTime = DateTime.Parse(date.Trim());
}
if (video.CreatedTime == null)
{
video.CreatedTime = DateTime.Now;
}
video.Caption = mainNode.SelectSingleNode(".//section[@class='ch9tab download']//div[@class='download']/div[2]")?
.InnerHtml;
video.UpdatedTime = video.CreatedTime;
return(video);
}
catch (Exception e)
{
Console.WriteLine(e.Message + e.StackTrace + url);
return(default);
// 临时补充遗漏
public C9Videos GetPageVideoByUrl(string fullUrl)
{
C9Videos video = new C9Videos();
try
{
var hw = new HtmlWeb();
// option获取InnerText，需要加以下设置
HtmlNode.ElementsFlags.Remove("option");
HtmlDocument htmlDoc = hw.Load(fullUrl);
HtmlNode mainNode = htmlDoc.DocumentNode.SelectSingleNode(".//main[@role='main']");
video.Duration = mainNode.SelectSingleNode(".//div[@class='playerContainer']//time[@class='caption']")?
.Attributes["datetime"]?.Value;
// 非视频,返回
if (video.Duration == null)
{
return(video);
}
video.SeriesTitle = mainNode.SelectSingleNode(".//div[@class='seriesTitle']//a")?.InnerText;
video.SeriesTitleUrl = mainNode.SelectSingleNode(".//div[@class='seriesTitle']//a")?.Attributes["href"]?.Value;
video.SourceUrl = fullUrl;
video.Title = mainNode.SelectSingleNode(".//div[@class='itemHead holder']//div[@class='container']//h1")?.InnerText;
video.ThumbnailUrl = mainNode.SelectSingleNode(".//div[@class='playerContainer']//a[@class='video']")
?.Attributes["style"]?.Value;
if (!IsNullOrEmpty(video.ThumbnailUrl))
{
Regex regex = new Regex(@".+\((.+)\);");
video.ThumbnailUrl = regex.Match(video.ThumbnailUrl).Groups[1].Value;
}
video.Author = mainNode.SelectSingleNode(".//div[@class='authors']")?.Descendants("a")?.Select(s => s.InnerText)
.ToArray().Join();
video.Language = mainNode.SelectSingleNode(".//div[@class='itemHead holder' and @dir='ltr']")?
.GetAttributeValue("lang", Empty);
video.Description = mainNode.SelectSingleNode(".//section[@class='ch9tab description']/div[@class='ch9tabContent']")
.InnerHtml;
var downloadUrls = mainNode.SelectNodes(".//section[@class='ch9tab download']//div[@class='download']//ul//li")?
.Select(s => new
{
text = s.Element("a").Attributes["download"]?.Value,
value = s.Element("a").Attributes["href"]?.Value
}).ToList();
Console.WriteLine(JsonConvert.SerializeObject(downloadUrls));
foreach (var downloadUrl in downloadUrls)
{
var downloadType = downloadUrl.text.ToLower().Trim();
if (downloadType.Contains(".mp3"))
{
video.Mp3Url = downloadUrl.value;
}
else if (downloadType.Contains("low.mp4"))
{
video.Mp4LowUrl = downloadUrl.value;
}
else if (downloadType.Contains("mid.mp4"))
{
video.Mp4MidUrl = downloadUrl.value;
}
else if (downloadType.Contains("high.mp4"))
{
video.Mp4HigUrl = downloadUrl.value;
}
}
video.Tags = mainNode
.SelectNodes(".//section[@class='ch9tab description']//div[@class='ch9tabContent']//div[@class='tags']//a")?
.Select(s => s.InnerText).ToArray()?.Join();
video.Views = 0;
video.CreatedTime = DateTime.Parse(mainNode.SelectSingleNode(".//time[@class='timeHelper']")?
.GetAttributeValue("datetime", Empty));
video.UpdatedTime = video.CreatedTime;
}
catch (Exception e)
{
System.Console.WriteLine(e.Message + e.StackTrace);
}
return(video);
}